Colonoscopy image analysis for polyp detection: A systematic review of existing approaches and opportunities

Abstract

Objective:

Colonoscopy is a diagnostic procedure using a flexible tube called a colonoscope with a camera to identify abnormalities in the large intestine and rectum, such as inflamed or swollen tissues, polyps, and cancer signs. It is crucial for early detection of colorectal cancer. However, analyzing colonoscopy images requires trained professionals, making it time-consuming and susceptible to errors. Advancements in machine learning have shown promising results in detecting polyps in colonoscopy images, improving efficiency. This paper provides a comprehensive overview of recent research in this field.

Methods and procedures:

This review uses the PRISMA (Preferred Items for Reporting Systematic Reviews and Meta-analyses) methodology, where an NLP (Natural Language Processing) toolkit, was used to search in several scientific databases, including IEEE Xplore, Springer, PubMed, Elsevier, and MDPI, published between 2010 and 2021, and related to colonoscopy detection based on image processing techniques.

Results:

This paper thoroughly analyzes the latest methods and prospects for polyp identification in colonoscopy pictures. Sixteen papers met the inclusion criteria, highlighting the need for automated system development and further research.

Clinical Impact:

The significance of the results lies in their ability to facilitate the creation of novel polyp identification techniques that medical professionals, trainees, and students may apply in near real-time.

Conclusion:

While every study that was given offers valuable insights into individual outcomes and methodology, no reports of clinical validation were made. A qualified individual’s validation is required for a method to be accepted. Even though the results are encouraging, the impact and applicability in actual situations are reduced in the absence of this phase.